No abstract

Humans continual learning (CL) ability is closely related to Stability Versus Plasticity Dilemma that describes how humans achieve ongoing learning capacity and preservation for learned information. The notion of CL has always been present in artificial intelligence (AI) since its births. This paper proposes a comprehensive review on CL. Different from previous reviews that mainly focus on the catastrophic forgetting phenomenon in CL, this paper surveys CL from a more macroscopic perspective based on the Stability Versus Plasticity mechansim. Analogous to biological counterpart, "smart" AI agents are supposed to i) remember previously learned information (information retrospection); ii) infer on new information continuously (information prospection:); iii) transfer useful information (information transfer), to achieve high-level CL. According to the taxonomy, evaluation metrics, algorithms, applications as well as some open issues are then introduced. Our main contributions concern i) recheck CL from the level of artificial general intelligence; ii) provide a detailed and extensive overview on CL topics; iii) present some novel ideas on the potential development of CL.

Generative adversarial networks (GAN) has been mainly used in the generation of natural images such as MNIST, CIFAR10 as well as Imagenet datasets and achieves satisfying generation results. However, GAN always fails in generating high quality high-resolution remote sensing images because remote sensing images are large in size and have various ground objects. To address this issue, a novel framework called High-Resolution PatchGAN (HRPGAN) is introduced in this paper. The structure of HRPGAN follows PatchGAN, but the batch normalization layers are removed and the ReLU activation is replaced by the SELU activation. In addition, a new loss function consisting of the adversarial loss, perceptual reconstruction loss and regularization loss is used in HRPGAN. Experiment results show that the proposed HRPGAN model generates the more diverse and lifelike images in HR remote sensing generation than Bicubic method and TGAN model.

Generative adversarial networks (GANs) suffer from catastrophic forgetting when learning multiple consecutive tasks. Parameter regularization methods that constrain the parameters of the new model in order to be close to the previous model through parameter importance are effective in overcoming forgetting. Many parameter regularization methods have been tried, but each of them is only suitable for limited types of neural networks. Aimed at GANs, this paper proposes a unified framework called Memory Protection GAN (MPGAN), in which many parametrization methods can be used to overcome forgetting. The proposed framework includes two modules: Protecting Weights in Generator and Controller. In order to incorporate parameter regularization methods into MPGAN, the Protecting Weights in Generator module encapsulates different parameter regularization methods into a "container", and consolidates the most important parameters in the generator through a parameter regularization method selected from the container. In order to differentiate tasks, the Controller module creates unique tags for the tasks. Another problem with existing parameter regularization methods is their low accuracy in measuring parameter importance. These methods always rely on the first derivative of the output function, and ignore the second derivative. To assess parameter importance more accurately, a new parameter regularization method called Second Derivative Preserver (SDP), which takes advantage of the second derivative of the output function, is designed into MPGAN. Experiments demonstrate that MPGAN is applicable to multiple parameter regularization methods and that SDP achieves high accuracy in parameter importance.INDEX TERMS Catastrophic forgetting, generative adversarial network, parameter regularization methods.

No abstract